RFR: 1912: Show priority for bugs in pull request body

Zhao Song zsong at openjdk.org
Tue May 23 16:56:47 UTC 2023

On Tue, 23 May 2023 12:57:42 GMT, Erik Joelsson <erikj at openjdk.org> wrote:

>> The initial purpose of this issue is to show priority for bugs in the issue list in the pull request body. However, with current pr bot, if the user changes the type of the issue or the priority of the issue in the JBS, the changes will not be reflected in the PR body unless there is something could trigger the CheckWorkItem(like user edits the PR title and user issues some commands).
>> As Erik.J suggested, we could maintain a map(from JBS issue to the PRs related with this issue) in memory, so if the issue gets updated, we could know which pr needs to be updated.
>> To query updated issues and handle them, a new bot called `IssueBot` is introduced. This bot would query for updated issues(exclude CSR and JEP) in JBS, and it will read the in memory map to know whether there is any pr needs to be updated. If so, it will generate `CheckWorkItem` for that pr.
>> So currently, there are two ways to generate `CheckWorkItem`. Updated issues and updated pull requests. Previously, in the `CheckWorkItem`, we would check the metadata of the pull request and if the metadata is up to date, `jcheck` would not be triggered.  Now, we could check the metadata of the pull request and the issues related to this pr, but it would be too expensive because if only the pull request is updated, we also need to fetch all the issues from JBS to just generate the metadata. Therefore, in this patch, metadata is split to two parts, one part for pull request and one part for issues. If the `CheckWorkItem` is spawned from an updated pull request, we will only check pr metadata. On the other hand, if the `CheckWorkItem` is spawned from an updated issue, we will only check issues metadata.
>> Besides, since the map is in-memory, so if the bot restarts, the map needs to be initialized. When bot restarts, a `CheckWorkItem` would be generated for each pr, so the initialization is in `CheckWorkItem`.
> bots/pr/src/main/java/org/openjdk/skara/bots/pr/CSRIssueWorkItem.java line 102:
>> 100:                 .flatMap(id -> bot.repositories().stream()
>> 101:                         .filter(r -> r.name().equals(id.split("#")[0]))
>> 102:                         .map(r -> r.pullRequest(id.split("#")[1]))
> I would suggest storing a better type than `String` in the issuePRMap so we don't have to parse it. Storing the full PullRequest object is a bit wasteful and could get confusing, but you could create a small `record` with just the repository name and pull request id.

It's a good idea. Will do it.


PR Review Comment: https://git.openjdk.org/skara/pull/1523#discussion_r1202711692

More information about the skara-dev mailing list